A ventilator management strategy that permits hypercapnia has several physiologic effects on gas exchange, cardiac output, and respiratory drive that may be beneficial. The increase in alveolar CO₂ that occurs during permissive hypercapnia increases CO₂ elimination for the same minute ventilation. This is because CO₂ elimination is governed by the following equation:

In this equation, K is a constant, VCO₂ is CO₂ elimination, PACO₂ is alveolar CO₂, and VA is alveolar ventilation. If the PACO₂ is allowed to rise and alveolar ventilation is kept constant, CO₂ elimination will increase. Therefore, a clinician can use lower tidal volumes and lower peak inspiratory pressures (decreasing alveolar ventilation) and allow the CO₂ to rise, yet maintain effective CO₂ removal. The hyperbolic relationship between PaCO₂ and minute ventilation permits effective CO₂ removal at a lower-than-expected amount of alveolar ventilation (Boynton and Hammond 1994). Thus, for example, if alveolar ventilation decreases by half, PaCO₂ will increase twofold but the increase in alveolar CO₂ tends to maintain a constant CO₂ elimination.

Aside from increasing the efficiency of CO₂ removal, hypercapnia (achieved with inspired CO₂ in animal studies) can improve ventilation-perfusion matching in the lung (Swenson et al. 1994; Brogan et al. 2004; Sinclair et al. 2006). In addition, hypercapnic acidosis permits more unloading of oxygen to the tissues due to the rightward shift of the oxygen dissociation curve (Bohr effect). Another physiologic advantage is stabilization (or increase) in respiratory drive, resulting in less apnea (Kondo et al. 2000) that can facilitate weaning an infant off the ventilator (Mariani et al. 1999; Søvik and Lossius 2004). Permissive hypercapnia also has beneficial effects related to cardiac function. Hypercapnia causes intracellular acidosis in the cardiac myocyte because of the high permeability to CO₂ (Laffey and Kavanagh 1999). In animals and humans, both therapeutic hypercapnia and permissive hypercapnia have been shown to decrease cardiac contractility but increase overall cardiac output (Walley et al. 1990; Weber et al. 2000). The increase in cardiac output is believed to be due to a fall in peripheral vascular resistance following hypercapnic acidosis (Cardenas et al. 1996). Furthermore, as tidal volume and peak inspiratory pressure are lowered, mean airway pressure decreases, which also may improve preload and cardiac output (Weiner et al. 1987; Gullberg et al. 1999, 2004).

Possible negative effects of hypercapnia on gas exchange include a small reduction in alveolar O₂ concentration caused by the increased alveolar CO₂. This can be overcome with slight increases in FiO₂ to increase alveolar oxygen concentration. Hypercapnic acidosis may also cause an increase in pulmonary vascular resistance in some infants although this has not been observed in clinical trials. In addition, while a higher PaCO₂ might be good, if the tidal volume is too low, the alveoli may collapse, resulting in right to left intrapulmonary shunting.

Hypercapnic acidosis also increases metabolic demand through an increase in work of breathing (Thome et al. 2006), but because oxygen delivery and tissue oxygenation is increased, this may not be clinically relevant. Hypercapnia may play a role in the development of retinopathy of prematurity (ROP) through retinal vessel vasodilatation, increased oxygenation, and subsequent formation of oxygen-derived free radicals (Holmes et al. 1998; Leduc et al. 2006). However, the randomized trials in neonates that tested for the presence of ROP or long-term visual outcomes reported no difference in the control versus the hypercapnia groups (Mariani et al. 1999; Thome et al. 2006; Carlo et al. 2002).

There is a particular concern in neonates of potential increased risk of IVH because of loss of cerebral autoregulation when exposed to high PaCO₂. However, this is not supported by the clinical trials (see later).

Permissive hypercapnia allows a clinician to manage a ventilated patient with the intent of reducing lung injury resulting from volutrauma. Furthermore, the resulting hypercapnic acidosis itself may be advantageous to the injured lung and to other organs. It is important to distinguish between permissive hypercapnia (the presence of hypercapnia in the setting of a ventilated patient being managed with low minute ventilation) and therapeutic hypercapnia (deliberate induction of hypercapnia by adding CO₂ to inspired gases, usually done experimentally) (Rotta and Steinhorn 2006). Each intervention produces hypercapnic acidosis, but therapeutic hypercapnia allows researchers to test experimentally the independent effects of hypercapnic acidosis. For example, in an in vivo rabbit model of high-tidal volume, ventilator-induced lung injury (25 mL/kg tidal volume in both experimental groups), therapeutic hypercapnia (PaCO₂ 80–100 mmHg versus 40 mmHg) attenuated various measures of lung injury (Sinclair et al. 2002). Similarly, in a rat model of endotoxin-induced lung injury in which lower volumes were used (4.5 mL/kg), therapeutic hypercapnia either before or after instillation of intratracheal endotoxin also reduced lung injury indices (Laffey et al. 2004). It is interesting to note that buffering hypercapnic acidosis to maintain “normal” pH may abrogate the protective effects of hypercapnic acidosis (Nichol et al. 2009). These studies demonstrate some of the potential benefits of hypercapnic acidosis per se in adult animal models in vivo. This is important because permissive hypercapnia, while decreasing the risk of lung injury due to mechanical reasons (lung stretch), also may be beneficial because of the additive effects of hypercapnia and of acidosis. Recent experimental research suggests that the beneficial effects of hypercapnic acidosis in the adult lung may translate to the preterm lung. Strand et al. (2003) showed that in preterm lambs with ventilator-induced lung injury, those exposed to high inspired PCO₂ had less lung injury by histopathology than control lambs. In newborn rats, hypercapnic acidosis had beneficial effects on pulmonary hypertension and vascular remodeling induced by chronic hypoxia (Kantores et al. 2006). Exposure to 10 % CO₂ attenuated increased oxidant stress induced by hypoxia, as quantified by 8-isoprostane content in the lung, and prevented upregulation of endothelin-1, a critical mediator of pulmonary vascular remodeling. Thus, CO₂ may have antioxidant properties on the lung and consequent modulating effects on the endothelin pathway. Vannucci et al. (1995) studied an immature rat model of cerebral hypoxia-ischemia, finding that mild hypercapnia was associated with less severe brain damage than normocapnic animals.

Li et al. (2006) investigated the effect of hypercapnia on overall growth characteristics, lung structure, and global gene expression in lungs of newborn mice exposed to 8 or 12 % CO₂ for 2 weeks. Their research demonstrated a substantial number of genes with altered expression in mice exposed to 8 % CO₂ but not in those exposed to 12 % CO₂. Of particular interest was an increase in transcription of the genes encoding surfactant-associated protein A and B (Sftp-A and Sftp-B). These proteins, expressed in bronchial epithelial and type II alveolar cells, may contribute to host defense mechanisms in the lung. Overall growth (body weight), hemoglobin concentration, and lung alveolar development were normal in both the hypercapnia and control groups. Recent preliminary data demonstrate that hypercapnia may accelerate alveolar development and attenuate hyperoxia-induced fibrosis in neonatal mice (Ryu et al. 2010; Das et al. 2009). These data suggest that hypercapnic acidosis may enhance host defense mechanisms through altered gene expression and promote alveolar development in the developing lung.

One of the benefits of hypercapnic acidosis demonstrated by in vivo animal studies is reduced neutrophil influx into the lung and reduced inflammatory cytokine production (Sinclair et al. 2002; Laffey et al. 2000, 2004). Such anti-inflammatory effects of hypercapnic acidosis also have been studied in vitro using neutrophils and endothelial cells. Hypercapnia can affect various neutrophil functions such as chemotaxis by altering the ability of neutrophils to regulate intracellular pH (O’Croinin et al. 2005). Hypercapnic acidosis also can alter neutrophil adherence to pulmonary endothelial cells by decreasing endothelial expression of the adhesion molecule ICAM-1 and of the chemoattractant interleukin-8 (Takeshita et al. 2003). Furthermore, hypercapnic acidosis can decrease lipopolysaccharide (LPS)-stimulated secretion of the acute proinflammatory cytokine tumor necrosis factor alpha in isolated rat macrophages (Lang et al. 2005a). The proposed mechanism of decreased expression of ICAM-1 and inflammatory cytokines appears to be related to decreased activation of nuclear factor kappa beta, a key early regulator of gene expression in the innate immune response (Watterberg et al. 1996). The ability of hypercapnic acidosis to modulate the innate immune response has important implications for lung development. Antenatal exposure to inflammation, such as chorioamnionitis, is a frequent cause of preterm birth and increases the risk of developing BPD (Watterberg et al. 1996; Jobe 2003). Furthermore, innate immune signaling can inhibit structural lung development in experimental animal models (Prince et al. 2005; Benjamin et al. 2007). Thus, hypercapnic acidosis also may be beneficial to the developing lung via its anti-inflammatory and immune-modulating capabilities. However, more experimental research is needed to determine how hypercapnic acidosis can be optimized to protect the preterm lung.

Even though most of the data indicate that hypercapnic acidosis is beneficial, some studies have reported neutral or negative results. Lang et al. (2005b) used intravenous LPS injection to mimic sepsis-related lung injury in adult rabbits and demonstrated that permissive hypercapnia contributed to lung injury via amplified oxidative injury. Other researchers have shown that hypercapnia may impair cell repair mechanisms after injury (Doerr et al. 2005) and induce injury to alveolar epithelial cells by potentiating tissue nitration (Lang et al. 2000). In an adult rabbit lung injury model using saline lavage to cause surfactant deficiency, therapeutic hypercapnia did not improve oxygenation or attenuate lung injury (edema, compliance, airway pressure, or cytokine production) after high-volume ventilation (12 mL/kg versus 5 mL/kg of tidal volume) (Rai et al. 2004). In eight adult patients with ARDS exposed to hypercapnia, there was a decrease in the Qp/Qs ratio, with deterioration in gas exchange (Feihl et al. 2000). Although a recent preliminary report suggested that therapeutic hypercapnia (5.5 % inspired CO₂) could protect the neonatal rat lung from hyperoxic injury (Masood et al. 2009), other investigators who used higher CO₂ concentrations (8 % CO₂) did not report this benefit (Das et al. 2009). Hypercapnia has been associated with increased risk of impaired cerebral blood flow autoregulation and intracranial hemorrhage in neonates (Kaiser et al. 2005). Kaiser et al. (2005) have examined the effects of endotracheal tube suctioning on cerebral autoregulation at various levels of PaCO₂. They found a relationship between acute PaCO₂ changes and cerebral blood flow. Marked hypercapnia can cause pressure-passive cerebral blood flow and increase risk of IVH. In another study, the same group suggested an association between severe hypercapnia in the first 72 h and IVH (Kaiser et al. 2006). It should be noted that the findings refer to fairly extreme degrees of hypercapnia, not what we typically feel to be appropriate target PaCO₂ levels, even in the context of permissive hypercapnia. In their study, the association of PaCO₂ and severe IVH involves very high levels (median 73.8, 25–75 % = 62.1, 84.7 for grade 3 IVH and median 73.9 25–75 % 65.4, 90.5 for grade 4 IVH). Even when a multivariate analysis was done, it is still not conclusive that there is a causative relationship, as opposed to an association of poor ventilation and severe IVH in the sickest babies. Furthermore, two large new clinical data suggest that severe hypocapnia and wide fluctuations in PaCO₂ in addition to hypercapnia were associated with increased risk of hemorrhage (Fabres et al. 2007; McKee et al. 2009) (see below). Nonetheless, there certainly is a plausible pathophysiologic mechanism and it is likely that very high PaCO₂ may be dangerous in the first 4 days of life. Other studies have shown that hypercapnic acidosis increases cerebral oxygen delivery (Hare et al. 2003) and that the CO₂-induced alterations in cerebral blood flow appear to be reversible (Hino et al. 2000).

A recent retrospective study of 849 infants weighing 1,250 g or less revealed that severe hypocapnia, severe hypercapnia, and wide fluctuations in PaCO₂ were associated with increased risk of hemorrhage (Fabres et al. 2007). The OR for IVH was 1.97 (1.23–3.15) if the maximal PaCO₂ was higher than 60 mmHg during the first 4 days of life and 2.51 (1.53–4.12) when the minimal PaCO₂ level was below 39 mmHg. Another retrospective study from the same group found that extreme fluctuations in PaCO₂ and higher max PaCO₂ were associated with worse neurodevelopmental outcomes in extremely low birth weight infants (McKee et al. 2009). They may indicate either a greater severity of illness or contribution of PaCO₂ to pathophysiology of adverse neurodevelopment. Most importantly, the randomized, controlled trials of permissive hypercapnia in neonates have not reported an increase in intracranial hemorrhage.

After the report by Avery et al. (1987) on 1987 showing the lowest rate of BPD in the center in which physicians practiced a permissive hypercapnia approach, several investigators have focused on the advantages of an early CPAP strategy. Avery’s publication was a retrospective analysis of 1,625 VLBW infants born between 1982 and 1984 in eight centers. Overall survival was 78–84 %, and BPD rate was significantly lower in Columbia, where 42 % of those infants had received mechanical ventilation and 37 % had been treated with early CPAP. The “Columbia approach,” as it was started to be called, included early nasal CPAP and acceptance of PaCO₂ levels as high as 60 mmHg. By that time, the same group was proposing a “gentler” approach to manage term neonates with PPHN (Wung et al. 1985), avoiding the widespread hyperventilation-hyperoxemic strategy. Unfortunately, the approach was not scientifically tested until recent years.

In 2000, Van Marter et al. (2000) published a cohort study including 441 VLBW infants (500–1,500 g) trying to identify variables associated with the development of BPD. The study compared outcomes in three clinical centers, and again, Columbia had a much lower rate of BPD, with lower use of surfactant and higher use of early CPAP. A multivariate analysis showed that intubation significantly increased the risk of BPD (OR 13.4, 95 % CI 5.9–30.7 on the day of birth).

In 2005, Ammari et al. (2005) shared the Columbia experience in a publication that intended to find variables associated with early CPAP failure in VLBW infants. They found the following ORs: GA (each week), 0.53 (0.43–0.65); PPV in the delivery room, 2.37 (1.02–5.52); AaPO₂ >180 mmHg, 2.91 (1.30–6.55); and severe RDS by X-ray, 6.42 (2.75–15). It was interesting that 92 % of infants between 1,000 and 1,250 g were successfully managed with early nasal CPAP, while the percentage was 33 % for those infants with BW below 700 g.

It is known that early surfactant administration reduces mortality and respiratory morbidity while the use of mechanical ventilation increases the risk for bronchopulmonary dysplasia. So, over the last few years, one of the controversies in the management of extremely preterm infants has been whether to intubate and give early surfactant or to start early nasal CPAP, avoiding intubation (Finer 2006).

Several observational studies, most with a “before-after” design, suggest that the use of early nasal CPAP reduces BPD rates (Dunn and Reilly 2003; Bohlin et al. 2008). Expert opinions abound on this issue, and even when (following Roman philosopher Seneca) opinions should be weighed and not just counted, they are the “lowest level of evidence.” Neonatologists have learned that the best way to protect neonates from “good ideas” is to test these ideas in randomized control trials.